American Mineralogist, Volume 71, pages 1337-1342, 1986
Evidencefor equilibrium conditionsduring the partitioning of
nickel betweenolivine and komatiite liquids
J. R. Buo.q,HN
U.S. Geological Survey, Box 25046,M5 424, Denver Federal Center, Denver, Colorado 80225, U.S.A.
Ansrnacr
Olivine-toJiquid partition coeftcients for Ni (D*,), calculated from Ni versus MgO
abundancevariations in komatiite seriesbasalts,compare favorably with experimentally
determined values if Ni variations in olivine-controlled basalts can be modeled with an
equation that assumes equilibrium between the entire olivine crystal and its coexisting
liquid. Thus, Ni abundancesin evolved komatiite seriesbasaltscan be calculatedby using
an equilibrium or Nernst crystallization equation and are independentof reasonablechoices
for the Ni and MgO content of the primary magma. Using published partition coefficients,
calculations of Ni and MgO abundancesin cumulate olivines in komatiite seriesbasalts
suggestthat Do,,valuesdetermined from natural basalt systemsare more geologicallyvalid,
i.e., producemore reasonableolivine compositions than do D",values from syntheticbasalt
systems.
INrnonucrroN
The difficulty in modeling the partit
the olivine-compatible element Ni in
(e.g.,komatiitesfisattributedtouncerta
olivine/liquid partition coefrcients for
vis, l97g; Irving, l97g; Elthon and I
though theseunCertaintiescan be relatr
conditions-that is, D", values obtainer
periments using natural basalts (Arnd
al.,1977)are diferent from thoseobta
syntheticbasalts(Hart and Davis, l9i
plained by stoichiometric control (Tak
tggt; Joo"r, lgg4)-significant differe
versusobservedNi abundancesresult fi
data to natural basalts.Empirically derj
partition coeftcients forNi (Cawhorn
can predict Ni variations in high-Mg(
nevertheless,thesevalues ditrer signifir
imental values for lower-MgO basalts.
Typical calculations for trace-elementpartitioning astn-" u fractional-crystallization processand emploi the
following equation:
ct/co: po-r,
(l)
Ni abundancein the basalt.Cawthorn and Mclver (1977)
J abundance data on an olivine-con'ic suite of Barberton Mountainland koalts from South Africa (Table l) and
Cer to calculate the required empirical
tition coefficientsfor Ni. Their calcuro possiblehigh-MgO primary magmas'
/o MgO and a c. : 2000 ppm Ni (model
o/oMgO and a c. : 1000ppm Ni (model
ersuslogC/c.generatedD",valuesfrom
veatselectedpoints.ElthonandRidley
the Cawthorn and Mclver values are
the experimentaldata on a synthetic
Davis, I 978; seeFig. l) and questioned
rf the latter data set to natural basaltsthe empirical values of Cawthorn and
from the experimental values derived
ltem (e'g', Arndt, 1977; seeFig' l) for
basalts'
The
of this paper is to further examine Ni
-purpose
versus MgO variations in komatiite seriesbasaltsand to
suggesla different equation for calculating Ni variations
in olivine-controlled magmas.Estimatesof Ni and MgO
compositions of cumulate olivines are used to establish
"best"_values'for olivine/liquid partition coefficientsfor
Ni and_are based on natural olivine compositions. AdditionalcomparisonsofpublishedNipartitioncoefficients
to empirically derived values, calculated from the proPosedcrystallization equation and Ni data on high-MgO
komatiite seriesbasaltswith well-characterizedfractionation sequences,confirm the proposed "best" values for
wherecis the abundanceofNi, for example,in an evolved
magma; c" is the abundanceof Ni in the parental magma;
f is ttre fraction of magma remaining; unOI it theiutt
partition coefficient foiNi. AttemptJto model Ni partiiioning among olivine-controlled basalts,using Equation
I and published D", values derived from experimental
studies,meet with limited successwhen large degreesof
olivine crystallization separatethe basalts.The calculated
, Th. -b"rt,,valuesreferto publishedvaluesthat appearto be
Ni content of a potential high-MgO parental basalt, for
tfr. -ort upp.opriateto basajticsystems,covering'tihewidest
example, is often significantly larger than the measured rangeof compoiition.
0003-004x/86/ trr2-r337s02.00
1337
r338
BUDAHN: Ni PARTITIONING BETWEEN OLIVINE AND KOMATIITE LIQUIDS
Table1. MgOandNi abundances
in komatiites
and
from
estimateddegreesof olivinefractionation
CawthornandMclver(1977)
Sampleno.
S3
sl1
s16
s28
ov3
ov4
ov5
SG3
SGSO
VB1
V84
v11
40J
53J
87J
88J
AUS
LV4
LV7
MgO
(wt%)
Ni
(ppm)
18.5
15.8
16.0
13.7
17.9
13.6
11.1
9.8
20.4
22.3
529
407
424
383
512
423
348
169
1102
935
337
179
107
205
1407
862
1203
1533
210
205
9.8
6.4
81
31.6
23.0
26.6
29.4
9.4
Olivine removed
(%)
Model 1
51
55
55
58
52
56
58
60
48
456
56
59
63
62
17C
434
Model 2
17
23
23
27
18
27
31
34
11
lln
Dto
z
>14
J
-12
z
o
10
24
34
38
36
2Et
25C
60
61
34
3s
Note: C : Cumulus-enrichedrock.
D",. The partitioning of MgO and FeO in komatiites is
also examined,and equationsfor Drro and Dr"ofor olivine
are presentedthat incorporate the resultsofJones (1984),
correlatingolivine/liquid Ni, MgO, and FeO partitioning.
EvrnnNcn FoR EeUILTBRIUMcRysrALLrzATroN
Presentedin Figure 2 and Table I are MgO and Ni
contentsof komatiites reported by Cawthorn and Mclver
from the Barberton Mountainland region. Also plotted in
Figure 2 are collected komatiite data of Beswick (1982).
Despite the fact that thesesamplesare from diferent geologic and geographicallocalities, Ni abundancesare quite
consistent among basalts having similar MgO contents
and exhibit a systematicvariation with respectto MgO.
As explained above, Cawthorn and Mclver applied the
data in Table I to a fractional-crystallizationequation and
derived D", values that moderately agree with experimental values for high-MgO (>130/0)basalts but differ
significantly from the experimentalvaluesdetermined for
lower-MgO basalts.However, when this same data set is
modeled assumingequilibrium crystallization, D*, values
are calculated that show better agreementto the experimental values over the entire MgO range.
In an equilibrium or Nernst crystallization equation,
36
9121518212421303336
MsO (PERCENT)
Fig. 1. Otivine/basalticliquid partition coefrcientsfor Ni
plottedversusMgOcontentofliquid. Literaturevaluesarefrom
Arndt (1977;brokenline),Hart andDavis(1978;line)andCawthorn and Mclver (1977;solid symbols).Opensymbolscalculated in this study employ komatiite data from Carthorn and
equation;model I is
Mclver and an equilibrium-crystallization
MgOand2000ppmNi primarymagma(circles),
basedon a 350/o
MgO and 1000ppm Ni primarymagma
andmodel2 on a 24o/o
(squares).
proximations. The calculatedD* values (plotted in Fig.
l) are inferred to correspondto the equilibrium value for
the evolved basalt. As an example,if sample OV5, which
contains9.80/oMgO and C: 169 ppm Ni (seeTable l),
was derived by 600/oolivine fractionation from a 350/o
MgO and 2000 ppm Ni (c.) primary magma, as proposed
by Cawthorn and Mclver via model 1, then 4 the fraction
of liquid remaining, equals 0.4 and the D* value for a
MgO basaltis given by (2000/169 - 0.4)/(l - 0.4)
9.80/o
or D*, : 19.I via an equilibrium approach.Alternatively,
if sample OV5 was derived from a 24o/oMgO,1000 ppm
Ni primary basaltby 340loolivine fractionation (model 2),
then.F equals0.66 and D*, equals(1000/169 - 0.66)/
(l - 0.66) or 15.5.As seenfrom Fig. l, the trend toward
large D*, values (> l2) for low-MgO basalts (<90/o)indicated by various experimental studies (Irving, 1978) is
supported by these calculations. Based on this general
agreementbetweenexperimental and ernpirical values, it
(2)
ct/c": 111P+ f'(l
D)1,
is suggestedthat an equilibrium-crystallization equation
where the variablesare defined as in Equation l. By rear- shouldbe applied to the partitioning of Ni betweenolivine
rangem€nt of Equation 2, D", values can be calculated and basaltic magmas. Comparisons of calculated to obfrom
served Ni versus MgO variations in several komatiite
validate this suggestion.There is, however,
(3) seriesbasalts
D: (cJc- n/$ - n.
strong supportive evidence that equilibrium conditions
The determination ofD*, valuesfrom Equation 3 employs are usually maintained between cumulate olivines and
Cawthbrn and Mclver's data and assumesthat the degregs their host basaltsduring crystallization. A diffusion study
of olivine crystallization they reported are reasonableap- of Hart (1981) indicatesthat Ni, Mg, and Fe, as well as
BUDAHN: Ni PARTITIONING BETWEEN OLIVINE AND KOMATIITE LIQUIDS
I339
1950
^Xt
dq
1500
x^
^x
x
1350
o
XX
gX
1200
> 1050
> 1050
L
o-
:
z
o
xo
soo
c
z
oo
9p
boo
r
@o
X
t
MsO(PERCENT)
Fig.2. Ni versusMgO abundance
datain komatiites.Data
represented
by o are from Cawthornand Mclver (1977);data culatedNi versusMgO liquid linesof descent
basedon an equirepresented
by arefrom Beswick(1982).
1750ppm Ni
equation,
a
32o/oMgO
and
librium-crystallization
"
primarymagma(asterisk),
dataof Arndt
andpartition-coefrcient
(1977;
line;
Eqs.
4,
6)
and
Hart and
and
broken
solid
circles
Co and Mn, will re-equilibrate betweenthe olivine phase
(1978;solidsquares
andsolidline;Eqs.5, 7).
and magmato relatively low temperatures(1075-1200'C) Davis
provided that the magma chamber is moderately large
(> 300 m in diameter) and the cooling rate is low (< 106 tional to the weight-ratio values for a given olivine and
t/m.y.). Such cooling rates would be applicable in the basalt composition) to give
magma chambersof most terrestrial basalts.
(6)
D-"o:29.34/(wto/oMgO) + 0.82
Appr,rclrroN oF EeuILrBRruM cRvSTALLIZATIoN To when combined with Equation 4 (Arndt) and
Fig.
3. ."-r""*"
"n**n:*;;:;,
NATURAL
; ;
BASALTS
In order to compare the Ni variations calculatedfrom
an equilibrium-crystallization model to those measured
in olivine-controlled basalts, appropriate olivine/liquid
partition coefficientsfor Ni and a reasonableestimate for
the degreeof olivine crystallization are required. Leastsquaresfitting of D*, versus MgO abundanceson two of
the more widely accepteddata sets for D", (presentedin
Fig. l) gives the following equations:
Dr,o:
MgO) + 1.12
31.6/(wto/o
(7)
when combined with Equation 5 (Hart and Davis).
Although alternative methods are available for determining the degree of olivine crystallization (e.g., variations in elementsincompatible with olivine, Fe-Mg partitioning), it is suggested that the equilibriumcrystallization Equation 2, the MgO abundancesof endmember basalts,and the Drro values from Equations 6 or
7 can be usedfor this estimate.Thus, the fraction of liquid
D*,: lls/(wto/oMgO) 2.1,
(4)
remaining, F, is given by (c"/C - D)/(l - D). In this
basedthe data reported by Arndt (1977) and,
approach,for eachvalue of F determined, C for Ni in the
evolved magmasis calculatedand compared to the meaD*,:l24/(trrtoloMgO) 0.9,
(5)
sured Ni abundancein the komatiite basalt series.
as given by Hart and Davis (1978).
The calculated liquid lines of descentfrom a primary
The study by Jones (1984) showing the linear correla- magma composition arbitrarily containing 320loMgO and
tion of molar Mg versus Fe, Mn, and Ni olivine/liquid
1750 ppm Ni are illustrated in Figure 3. As seenfrom the
partition coefficients,coupled with the results suggested figure, both the Arndt and the Hart and Davis data sets
by this study and the difi.rsion study of Hart (1981), imare internally consistentand produce the Ni versus MgO
plies that MgO variations can also be modeled with an trends observed in komatiites when applied to an equiequilibrium equation. The equation presentedby Jones librium-crystallization equation. However, evidencethat
conelating molar NiDvalueswith molar MeDvalues(ND :
the D", values of Arndt are more geologically valid is
3.92-"D - 5.30) are combined with Equations 4 and 5 found in the calculated cumulate-olivine compositions.
(since molar-ratio partition coefficients will be proporEstimatesof Ni and MgO abundancesin cumulate ol-
I 340
BUDAHN: Ni PARTITIONING BETWEEN OLIVINE AND KOMATIITE LIQUIDS
ivines derived from komatiite seriesbasaltsare calculated
by multiplying the Ni and MgO contents of the basalt
times the respectiveolivine/liquid partition coefficients.
The cumulate-olivine compositions calculatedfor the selected intermediate basaltsin Figure 3 (solid circles and
squares)are presentedin Table 2. If mantle-derived olivines containing -3800 ppm Ni representan upper limit
to the Ni content in cumulate olivines, then the data set
of Arndt appearsto be the better one. The calculatedNi
content for an olivine crystallizing from the 27o/oMgO
intermediatebasalt in this example is 2. 16 x 1390 or
-3000 ppm Ni on the basisof the Arndt data set,whereas
the Hart and Davis values indicate that such olivines
contain 3.69 x 1270or -4700 ppm Ni. Additional support for the D", values of Arndt is also found in the calculated MgO content of cumulate olivines. The MgO
abundances(47-52o/o)estimated for olivines crystallizing
from the high-MgO basalts (>22o/o)by using the Dr*
partition coefficientsderived from Arndt's data are more
comparableto observedolivine compositions (Deer et al.,
1966) than the MgO abundances(56-620/o)determined
by using the Hart and Davis data.
Empirical partition coefficientsfor N| MgO, FeO, and
MnO are also examined in komatiite series basalts for
which independentmethods are available for determining
the degreeof olivine crystallization. For example, an approximate value for the fraction of liquid remaining in
the evolution of l0 Barberton aphyric komatiite series
basalts(Table 3) reported by Smith and Erlank (1982) is
obtained from the mean ratios ofthe olivine-incompatible
elementsTiOr, AlrOr, and CaO betweena primary magma
and an evolved basalt sinceC/c" x' l/F when D = 0. The
primary magmaassumedfor this komatiite seriescontains
0.26o/oTiOr,2.9o/o
AlrOr, 10.50/o
FeO, 0.170lo
MnO, 35.00/o
MgO,4.7o/oCaO, and 2050 ppm Ni and is similar to the
composition proposed by Smith and Erlank. The individual and mean ,F values for each of the basalts are
presented in Table 3. In this approach, the calculated
empirical D"i, Dueo,Dr"o, and Drno valuesassumingequilibrium crystallization are compared to the experimental
D values derived from the Arndt data set (Eqs. 4 and 6)
and the equations presentedby Jones (1984) correlating
FeO and MnO partition coefficientswith Drro:
Do.o: 0.277D*"o+ 0.126
(8)
Du"o : 0.225DME,
+ 0.073.
(g)
and
As indicated in Table 3, the empirical olivine/liquid D,,
values for these high-MgO basalts (>260/o)compare favorably with the experimental values of Arndt. The empirical olivine/liquid D*"o, Dr"o,and D*"ovalues are also
similar to the values calculated by using Equations 6, 8,
and 9. The disagreementbetween the empirical and experimentalD*,valuesfor samplesHSS-95,HSS-88A,and
HSS-88C, coupled with the variation in the estimated,F
values, indicates the participation of an additional crystallizing phasein the evolution ofthese three basalts.The
higher .F values calculated from TiO, and AlrO. abun-
in intermediate
MgOandNi abundances
Table2. Estimated
magmasevolvedlroma 32h MgO,1750ppmNi
orimarvbasalt
Cumulate
olivines
lntermediate
magmas
HD
MgO
(%)
Ni
(ppm)
27
22
17
12
7
27
22
1270
840
510
290
135
1420
94s
t/
coc
12
7
305
135
F
D,no
D",
MgO
(Y"l
Ni
(ppm)
0.86
0.71
0.55
0.40
0.23
0.80
0.60
0.43
0.27
0.11
2.29
2.56
2.98
3.76
5.64
1.91
2.15
2.55
3.27
5.01
3.69
474
6.39
9.43
16.8
2.16
3.13
4.66
7.48
14.3
61.8
56.3
50.7
45.1
39.5
51.6
47.3
43.4
39.2
35.1
4690
3980
3265
2725
2235
3070
2960
2630
2280
1930
and D"' values of Hart
A/ote.'Valuesderived from Equation 2 and DMso
and Davis(HD, 1978)and Arndt (A, 1977).F: Fractionof liquidremainIng.
dances compared to CaO suggestspinel crystallization,
since spinel has a higher affinity for Ti and Al than for
Ca (Deer et al., 1966).
Finally, the olivine-incompatible rare-earth elements
(REE) also provide a method for calculating the fraction
of liquid remaining following olivine crystallization. For
the Tipasjarvi komatiite seriesbasalts(Table 4) with strong
light REE{epleted patterns, Auvray et al. (1982) proposeda crystallization model between a primary magna
(sample5834) containing2T.2o/o
MgO and 1100 ppm Ni
with averageheavy REE =4x (times chondritic values)
and sample S83l (15.90loMgO, 650 ppm Ni, and
HREE t6 x) and samplesS8l8 and 5828 (average:7.4o/o
MgO, : I l0 ppm Ni, and HREE = I1.8 x). Althoueh clinopyroxene(cpx) and/or spinel are reported as additional
crystallization phasesin the evolution of the 7.4o/oMgO
basalt, the value determined for -F based on diferences
in HREE abundancesis virtually identical to the value
calculated from MgO abundancesand partition coefficients, Equation 6, and an equilibrium equation. This
agreementsuggeststhat olivine is the dominant cumulate
phasein the evolution of these basalts.Assuming, however, that the proportion of phasescrystallizing is -800/o
spinel,a value of -F: 0.32 is
olivine * l0o/ocpx * 100/o
required on the basis of partitioning of the HREE. The
MgO and Ni contents calculated for the evolved basalt
from Equation2, a value of F: 0.32, D'.o: 3.0 for cpx
and spinel, D", : 8 for cpx, and D*, : l3 for spinel (Irving,
1978)are 7.9o/oand 114 ppm, respectively.Theseabundances are comparable to the measured abundancesof
7.4o/oMgO and I l0 ppm Ni and support an equilibrium
model. On the basis of the differencebetween the calculated and measuredNi abundancesfor sample 5831, it
can be suggestedthat this sampleis not comagmaticwith
the proposed primary magma. The difference in chondrite-normalized Ce/HREE ratios between sample S83l
(0.58) and those of both the primary magma (0.70) and
the 7.4o/oMgO basalt (0.70) supports this conclusion.
1341
BUDAHN: Ni PARTITIONING BETWEEN OLIVINE AND KOMATIITE LIQUIDS
Tabfe 3. Comparisonof experimentaland empirical Dr"o, Drno,D"no,and D* values based on Barberton Mountainland
komatiites*and an equilibrium-crystalliztion
equation
HSS
109
o.28
3.1
11.2
0.17
329
4.8
2001
Tio,
Alro3
FeO
MnO
Mgo
CaO
Ni
Tio,
Al,o3
CaO
Mean
Dr.o
D"no
D"no
0.93
0.93
0.98
0.95
(1.0)
2.3
1.5
0.60
0.46
1.7
1.4
D*o
D"no
D"no
n
HSS
15
0.33
3.4
11.1
0.19
31.5
5.7
2095
0.79
0.85
0.82
0.82
0.70
0.42
16
0.61
o.47
1.75
| .cc
HSS
14
HSS
87
0.35
3.8
11.2
0.20
30.8
6.0
1833
0.35
3.6
12.2
0.20
30.6
5.5
1579
0.38
4.O
11.2
0.'t7
29.4
0.41
4.5
1 1. 7
0.20
27.0
7.9
6.5
1696
1375
HSS
95
HSS
88A
0.39
4.2
11.7
0.20
26.1
8.3
1426
0.27
4.9
12.5
0.21
25.7
7.4
1499
0.43
4.6
11.7
0.22
25.0
9.1
1305
0.39
4.1
10.7
0.18
24.1
10.1
1145
0.67
0.69
0.57
0.64
0.59
0.64
0.61
0.60
0.63
0.52
0.58
0.67
0.71
0.47
0.61
o.72
0.58
1.9
2.2
0.59
0.51
1.9
1.9
0.76
0.46
1.95
2.36
0.95
0.86
2.16
3.0
0.66
0.51
1.94
2.3
0.68
0.51
1.96
2.4
0.69
0.52
2.0
2.5
0.70
0.53
2.0
2.7
F(fractionof liquidremaining)valuesbasedon:
0.63
o.74
o.74
0 68
u./o
0.64
0.80
0.73
0.78
0.72
0.59
0.85
0.62
0.76
0.80
0.71
Empiricalpartition coefficients:"
o.74
0.38
1.6
1.5
0.62
o47
1.8
1.63
(0.3)
(0.25)
1.7
z.c
0.73
0.61
1.8
2.3
0.78
(1.0)
1.66
1.7
Experimentalpartition coefficients:f
0.65
0.62
0.635
o.47
1.8
1.66
0.50
1.9
2.16
0.48
18
1.81
HSS
88C
HSS
90
HSS
89
HSS
523
- Smith and Erlank(1982).Abundancesin percent,except Ni (ppm).Primarymagma:TiO,: 0.26, AbO3:2.9, FeO : 10.5,MnO : 0.17, MgO:
35.0. CaO : 4.7. Ni : 2050.
-- Calculatedfrom mean Fvalue and an equlibrium-crystallization
equation(Eq.2).
t Calculatedfrom Equations4, 6, 8, and 9.
CoNcr-uorNc REMARKS
The difficulty in modeling Ni abundancesin olivinecontrolled basaltic liquids may be largely a problem of
selectingthe appropriate crystallization equation rather
than the appropriate partition coemcients.Nevertheless,
Ni, MgO, and FeO abundancesin komatiite seriesbasalts
and cumulate olivines appearto be more accuratelymodeled using an equilibrium-crystallization equation and
experimental olivine/liquid partition coefrcients that are
derived from natural basalt systems.If an equilibriumcrystallization equation is correct for modeling other ol-
Table4. Comparison
of estimateddegreesof olivine
crystallization
andNi abundances
in
(Finland)
Tipasjarvi
komatiites
s818,
s831 5828
MgO (wto/")
Ni (ppm)
HREEN
(Ce/HREE)"
Duqo
D",
F
MgO (wt%)
Ni (ppm)
Ni (ppm)
15.9
7. 4
110
650
6.0 11.8
0.58 0.70
2.67 4.78
13.4
5.1
0.67 0.34
7.4
17.2
440
111
0 57
0.29
112
398
ivine-controlled basalts, then unique primary magma
compositions cannot be calculated using Ni abundance
variations sinceit can be shownthat MgO versusNi liquid
lines of descent from primary basalts containing l0 to
250loMgO are virtually identical. Petrogeneticmodels that
employ a fractional-crystallization equation to constrain
primary magmacompositionsbasedon MgO andNi contents may require re-evaluation. Although quantitative
in primaryand
of MgOandNi abundances
Table5. Estimates
in cumulate
magmasan-dNi abundances
intermediate
olrvrnes
Primary
magmas
tc.l
MgO
(wt%)
Ni
(ppm)
MgO
(wt%)
Ni
(ppm)
F
("/")
Cumulate
olivines
[D"' x c']
Ni (ppm)
10.8
350
13.6
540
16.4
725
19.2
910
10.0
8.0
12.0
10.0
8.0
14.8
12.5
10.3
8.0
17.5
15.1
12.8
10.4
8.0
280
165
390
260
165
565
390
260
165
750
530
375
255
165
97
90-94
87
80.94
86
78
70-94
84
76
68
60-.
2630
2025
2910
2430
2025
3225
2775
2370
2025
3385
2910
2600
2300
2025
Method of calculation
Eq. 6
Eq. 4
HREE ratios (primary/evolved)
Eq. 2 and Ffrom HREE ratios
Eq. 2 and Ffrom HREE ratios
Eqs.2 and 6 and MgO abundances
Eq. 2 and Ffrom MgO contents
Nofe; Calculationsuse REE and MgO data and a primarymagmawith
27.2o/oM}O,1100ppm Ni, and HREE : 4 x chondrites.F: Fractionof
liquidremaining.
Intermediatemagmas
lc'l
' F: percentageof liquidremainrng.
** AverageKilaueanlow-Mgo basalt: 8% MgO, 165 ppm Ni (Leeman
et al., 1980).
t342
BUDAHN: Ni PARTITIONING BETWEEN OLIVINE AND KOMATIITE LIQUIDS
geneticrelations in komatiite suites.In N.T. Arndt et al., eds.
Komatiites, 283-308. Allen and Unwin, London.
Bickle,M.J., Ford,C.E.,andNisbet,E.G.(1977)Thepetrogenesis
of peridotitic komatiites: Evidencefrom high-pressuremelting
experiments.Earth and Planetary ScienceLetters, 37,97-106.
Cawthorn, R.G., and Mclver, J.R. (1977) Nickel in komatiites.
Nature, 266, 7 16-718.
Deer, W.A., Howie, R.A., and Zussman,J. (1966)An introduction to the rock-forming minerals. Wiley, New York.
Elthon, Don, and Ridley, W.I. (1979) Comments on "The partitioning of nickel between olivine and silicate melts by S.R.
Hart and K.E. Davis." Earth and Planetary ScienceIrtters,
44, 162-164.
Gunn, B.M. (1971) Trace element partitioning during olivine
fractionation of Hawaiian basalt. Chemical Geology, 8, l-13.
Hart, S.R. (1981) Diftrsion compensation in natural silicates.
Geochimica et Cosmochimica Acta, 45,279-292.
Hart, S.R., and Davis, K.E. (1978) Nickel partitioning between
olivine and silicate melt. Earth and Planetary Science Irtters,
40,203-219.
Hebert, R., Bideau, D., and Hekinian, R. (1983) Ultramafic and
mafic rocks from the Garret transform fault near 13o30'Son
the East Pacific Rise: Igneous petrology. Earth and Planetary
ScienceLetters. 65. 107-125.
Irving, A.J. (1978) A review of experimental studies of crystaV
liquid trace elementpartitioning. Geochimica et Cosmochimica Acta, 42,743-770.
Jones,J.H. (1984) Temperature- and pressure-independentcorrelations of olivine/liquid partition coefrcients and their application to trace element partitioning. Contributions to Mineralogy and Petrology, 88, 126-132.
Leeman, W.P., and Scheidegger,K.F. (1977) Olivine/liquid distribution coefficientsand a test for crystal-liquid equilibrium.
AcxnowlnocMENTS
Earth and Planetary Sciencektters, 35,247-257.
Thismanuscript
wasgreatlyimprovedbythereviewcomments Leeman, W.P., Budahn, J.R., Gerlach, D.C., Smith, D.R., and
of K. L. Nash,R. A. Zielinski,K. Schulz,S.R. Hart,J. H. Jones, Powell, D.N. (1980) Origin of Hawaiian tholeiites: Trace element constraints. American Journal of Science,2804, 794rewiewer.
andan anonymous
81 9 .
Smith, H.S., and Erlank, A.J. (1982) Geochemistry and petroRnrBnnNcBs
genesisofkomatiites from the Barbertongreenstonebelt, South
Arndt, N.T. (1977)Partitioningof nickel betweenolivine and
Africa. In N.T. Arndt et al.. Eds. Komatiites, 347-397. Allen
Institutionof
ultrabasicandbasickomatiiteliquids.Carnegie
and Unwin, London.
Washington
YearBook,76, 553-557.
Takahashi,Eiichi, and Irvine, T.N. (1981)Stoichiometriccontrol
Auvray,B., Blais,S., John,B.M., and Piquet,D. (1982)Koof crystal/liquid single-componentpartition coefrcients. Geomatiitesand the komatiitic seriesof the Finnishgreenstone chimica CosmochimicaActa,45, 1181-l 185.
predictions remain diftcult, an alternative approach that
employs the measuredNi content in phenocryst olivines
is suggestedto constrain the compositions of primary
magmas. For example, olivines ranging in composition
from2525 ppm Ni (Gunn, l97l) to ashigh as 3400-3600
ppm Ni (Leeman and Scheidegger,1977) have been reported in 6-100/oMgO Kilauean basalts.On the basis of
averageMgO and Ni contents of 80/oand 165 ppm, respectively, in Kilauean lavas (Leeman et al., 1980), the
MgO and Ni contents of possible primary magmas, intermediate magmas,and olivine-cumulate compositions
are calculatedfrom Equations2,4, and 6. The results for
hypothetical primary magmascontaining 10.80/o
MgO ( I 0o/o
olivine fractionation), 13.60/oMgO (200lo),16.40loMgO
(300/o),and 19.2o/oMgO (400/o)are presentedin Table 5.
These calculations suggestthat the olivines found in Kilauean magmas containing 2525 ppm Ni are capable of
being derived from a 100/oMgO parent magma, whereas
the olivines containing 3400-3600 ppm Ni must have
beenderived from a primary or near-primary magmawith
> 190/o
MgO. Similar calculationsbasedon an 8.70loMgO
and 150 ppm Ni East Pacific Rise basalt (Hebert et al.,
1983)indicate that megacrystolivines with 2200 ppm Ni
werederived from a l0o/oMgO basaltand the Fon,olivines
containing 2800 ppm Ni were derived from an -130lo
MgO primary magma.
belts.In N.T. Arndt et al., eds.Komatiites,131-146.Allen
andUnwin, London.
MrNuscnlpr RECETVED
J,c.NuA.nv2, 1986
A.E.(1982)Somegeochemical
Beswick,
aspects
of alterationand MlNuscnrpr AcCEPTED
Jurv 8, 1986